sudhanshu_kasewa

Karma: 73

sudhanshu_kasewa 9 Apr 2026 22:05 UTC
1 point
0
on: My unsupervised elicitation challenge
Also chatty. Attempt 2
Α γράμμα ἐστίν. Α καὶ Β γράμματα εἰσιν. Α, Β, καὶ Γ τρία Ἑλληνικὰ γράμματά εἰσιν. Καὶ Π Ἑλληνικὸν γράμμα ἐστίν, οὐ Λατινικόν. C Λατινικὸν γράμμα ἐστίν, οὐχ Ἑλληνικόν. Β οὐ φωνῆεν, ἀλλὰ σύμφωνον ἐστιν. Β καὶ Γ οὐ φωνήεντα, ἀλλὰ σύμφωνα εἰσιν. Β οὐ μικρὸν γράμμα ἐστίν, ἀλλὰ κεφαλαῖον. β οὐ κεφαλαῖον, ἀλλὰ μικρὸν γράμμα ἐστίν. Ω = ὦ μέγα, Ο = ὂ μικρόν. ΑΙ Ἑλληνικὴ δίφθογγος ἐστιν. ΑΙ καὶ ΕΙ Ἑλληνικαὶ δίφθογγοι εἰσιν. Α’ δίφθογγος οὐκ ἔστιν, ἀλλ’ ἀριθμός. Α’ καὶ Β’ ἀριθμοί εἰσιν. «Ἀπολλώνιος» κύριον ὄνομα ἐστιν. «Ἀπολλώνιος» καὶ «Ἑλένη» κύρια ὀνόματα εἰσιν. «Ἀπολλώνιος» ἀρσενικὸν ὄνομά ἐστιν (♂). «Ἑλένη» θηλυκὸν ὄνομά ἐστιν (♀). «Salve» Λατινικὴ λέξις ἐστίν, οὐχ Ἑλληνική. «Salve» καὶ «lingua» δύο Λατινικαὶ λέξεις εἰσίν. «Χαῖρε», «γλῶσσα», καὶ «ἀριθμός» τρεῖς Ἑλληνικαὶ λέξεις εἰσίν. The two places most worth double-checking against your worksheet are: Α’ / Β’ = ἀριθμός / ἀριθμοί τρεῖς in the last line, because λέξεις is feminine plural Also, οὐδέτερον from the word bank appears to be unused.

sudhanshu_kasewa 9 Apr 2026 21:58 UTC
2 points
0
on: My unsupervised elicitation challenge
Edit—chatgpt, not Claude. Also, apologies about how it renders, it looks fine (with line breaks) while I type this. Edit 2 -- oh, formatting is better now.
Attempt 1
Α γράμμα ἐστίν.
Α καὶ Β γράμματα εἰσιν.
Α, Β, καὶ Γ τρία Ἑλληνικὰ γράμματά εἰσιν.
Καὶ Π Ἑλληνικὸν γράμμα ἐστίν, οὐ Λατινικόν.
C Λατινικὸν γράμμα ἐστίν, οὐχ Ἑλληνικόν.
Β οὐ φωνῆεν, ἀλλὰ σύμφωνον ἐστιν.
Β καὶ Γ οὐ φωνήεντα, ἀλλὰ σύμφωνα εἰσιν.
Β οὐ μικρὸν γράμμα ἐστίν, ἀλλὰ κεφαλαῖον.
β οὐ κεφαλαῖον, ἀλλὰ μικρὸν γράμμα ἐστίν.
Ω = ὦ μέγα, Ο = ὂ μικρόν.
ΑΙ Ἑλληνικὴ δίφθογγος ἐστιν.
ΑΙ καὶ ΕΙ Ἑλληνικαὶ δίφθογγοι εἰσιν.
Α’ δίφθογγος οὐκ ἔστιν, ἀλλ’ γράμμα.
Α’ καὶ Β’ δύο εἰσιν.
«Ἀπολλώνιος» κύριον ὄνομα ἐστιν.
«Ἀπολλώνιος» καὶ «Ἑλένη» κύρια ὀνόματα εἰσιν.
«Ἀπολλώνιος» ἀρσενικὸν ὄνομά ἐστιν (♂).
«Ἑλένη» θηλυκὸν ὄνομά ἐστιν (♀).
«Salve» Λατινικὴ λέξις ἐστίν, οὐχ Ἑλληνική.
«Salve» καὶ «lingua» δύο Λατινικαὶ λέξεις εἰσίν.
«Χαῖρε», «γλῶσσα», καὶ «ἀριθμός» τρεῖς Ἑλληνικαὶ λέξεις εἰσίν.
One line looks slightly odd as printed:
Α’ καὶ Β’ δύο εἰσιν.
That is the best fit from the word bank, but in fully natural Greek you would expect something like δύο γράμματα.

sudhanshu_kasewa 8 Mar 2024 12:13 UTC
2 points
0
in reply to: Kabir Kumar’s comment on: No Clickbait—Misalignment Database
No, this is not something I can undertake—however, the effort itself need not be very complicated. You’ve already got a list of Misalignment types in the form: create a google doc with definitions/descriptions of each of these, and put a link to that doc in this question.

sudhanshu_kasewa 23 Feb 2024 16:27 UTC
2 points
0
on: No Clickbait—Misalignment Database
It might be worth (someone) writing out what is meant by each kind of misalignment category, as used in the db. Objective misalignment, specific gaming, value misalignment all seem overlapping, and I’m not at all sure what physical misalignment is supposed to be pointing to.

sudhanshu_kasewa 28 Aug 2023 14:42 UTC
1 point
0
on: Who can most reduce X-Risk?
MIRI

sudhanshu_kasewa 28 Aug 2023 14:42 UTC
1 point
0
on: Who can most reduce X-Risk?
Anthropic

sudhanshu_kasewa 28 Aug 2023 14:41 UTC
1 point
0
on: Who can most reduce X-Risk?
The EU parliament

sudhanshu_kasewa 28 Aug 2023 14:41 UTC
1 point
0
on: Who can most reduce X-Risk?
The UK government

sudhanshu_kasewa 28 Aug 2023 14:41 UTC
1 point
0
on: Who can most reduce X-Risk?
Vladimir Putin

sudhanshu_kasewa 28 Aug 2023 14:41 UTC
2 points
0
on: Who can most reduce X-Risk?
The Chinese Communist Party

sudhanshu_kasewa 28 Aug 2023 14:40 UTC
1 point
0
on: Who can most reduce X-Risk?
The American Public

sudhanshu_kasewa 28 Aug 2023 14:40 UTC
2 points
0
on: Who can most reduce X-Risk?
Joe Biden

sudhanshu_kasewa 28 Aug 2023 14:40 UTC
1 point
0
on: Who can most reduce X-Risk?
Greg Brockman/Sam Altman

sudhanshu_kasewa 28 Aug 2023 14:39 UTC
1 point
0
on: Who can most reduce X-Risk?
Demis Hassabis

sudhanshu_kasewa 28 Aug 2023 14:39 UTC
1 point
0
on: Who can most reduce X-Risk?
Elon Musk

sudhanshu_kasewa 28 Aug 2023 14:38 UTC
0 points
0
on: Who can most reduce X-Risk?
Mark Zuckerberg

[Question] Who can most reduce X-Risk?

sudhanshu_kasewa28 Aug 2023 14:38 UTC

1 point

12 comments1 min readLW link

sudhanshu_kasewa 28 Jul 2023 11:35 UTC
2 points
0
on: Neuronpedia—AI Safety Game
Very cool. Thanks for putting this together.
Half-baked, possibly off-topic: I wonder if there’s some data-collection that can be used to train out polysemi from a model by fine-tuning.
e.g.:
- Show 3 examples (just like in this game), and have the user pick the odd-one-out
  - User can say “they are all the same”, if so, remove one at random, and replace with a new example
- Tag the (neuron, positive example) pairs with (numerical value) label 1, the odd-one-out with 0
- Fine-tune with next-word-prediction and an auxilliary loss using this new collected dataset
  - Can probably use some automated (e.g. semantic similarity) labelling method to cluster labelled+unlabelled instances, to increase the size of the dataset
Neuronpedia interface/codebase could probably be forked to do this kind of data collection very easily.

sudhanshu_kasewa 9 Feb 2022 10:30 UTC
2 points
0
on: Prizes for ELK proposals
Dumb question alert:
In the appendix “Details for penalizing depending on “downstream” variables”, I’m not able to wrap my head around what we can expect the reporter to learn—if anything at all—seeing that it has no dependency on the inputs (elsewhere it is dependent on z sampled from the posterior).
Specifically, the only call to the reporter (in the function reporter_loss in this section) contains no information (about before, action, after) from the predictor at all:
answer = reporter(question, ε, θ_reporter)
(unless “question” includes some context from the current (before, action, after) being considered, which I’m assuming is not the case)
My dumb question then is:
-- Why would this reporter be performant in any way?
My reasoning: For a given question Q (say, “Is the diamond in the room?”) we might have some answers of “Yes” and some of “No” in the dataset, but without the context, we’re essentially training the reporter to map noise that is uncorrelated with/independent of the context to the answer; essentially, for a fixed question Q and fixed realization of the noise RV, the reporter will be uniformly uncertain (or well, it will mirror the statistics in the data) about the value of the answer. Since the noise is independent/uncorrelated, this would be true for every noise value.

sudhanshu_kasewa 4 Feb 2022 16:02 UTC
1 point
0
on: Prizes for ELK proposals
Naive thought #2618281828:
Could asking counterfactual questions be a potentially useful strategy to bias the reporter to be a direct translator rather than a human simulator?
Concretely, consider a tuple (v, a, v’), where v := ‘before’ video, a := ‘action’ selected by SmartVault or augmented-human or whatever, and v’ := ‘after’ video.
Then, for some new action a’, ask the question:
- “Given (v, a, v’), if action a’ was taken, is the diamond in the room?”
(How we collect such data is unclear but doesn’t seem obviously intractable.)
I think there’s some value here:
- Answering such a question might not require computation concerning a and v’ ; if we see these computations being used, we might derive more value from regularizers that penalize downstream variables (which now includes the nodes close to a)
- This might also force the reporter to essentially model (or compress but not indefinitely) the predictor; the reporter now has both a compressed predictor Bayes’ net and a human Bayes’ net. If we can be confident that the compressed predictor BN is much smaller than the human BN, then doing direct translation within the reporter, i.e. compressed predictor BN inference + translation + read off from human BN might be less expensive than the human simulator alternative, i.e. compressed predictor BN inference + ‘translation’/bridging computation + human BN inference.
  - We might find ways of being confident that the compressed predictor BN is small (e.g. by adding decoders at every layer of the reporter that reconstruct v, a or v’ and heavily penalizing later-layer decoders)

sudhanshu_kasewa

[Question] Who can most re­duce X-Risk?

[Question] Who can most reduce X-Risk?